Parenting Stress Mediates the Association Between Negative Affectivity and Harsh Parenting: A Longitudinal Dyadic Analysis

Abstract

The current study examined parenting stress (disaggregated into personal distress and childrearing stress) at 12 months postpartum as a mediator of the longitudinal association between parental negative affectivity at 6 months postpartum and harsh parenting at 3 years postpartum for first-time parents with a child transitioning from late toddlerhood to the early preschool years. Analyses were conducted using Mediation for Actor Partner Interdependence Modeling in a sample of 164 couples who participated in a randomized controlled trial of a universal, couple-based transition to parenthood program. There were indirect actor effects of negative affect on a parent’s own harsh parenting through both dimensions of parenting stress, with a stronger mediating effect for personal distress than childrearing stress. There were also indirect partner effects of negative affect on one’s partner’s harsh parenting through the partner’s parenting stress, with a stronger indirect partner effect from mothers’ negative affect to fathers’ harsh parenting than vice versa. Specifically, the mediating effect of personal distress was found for both mothers and fathers, whereas the mediating effect of childrearing stress was found from mothers’ negative affect to fathers’ harsh parenting only. Findings highlight the importance of a dyadic approach in examining the longitudinal association between negative affect and harsh parenting and suggest that reducing parenting stress in the first year postpartum may decrease the risk of future harsh parenting among couples in which one or both partners experience negative affectivity.

Harsh parenting is common during the early years of parenthood (Kim, Pears, Fisher, Connelly, & Landsverk, 2010), and numerous studies have documented its adverse effects on child outcomes (e.g. Gershoff, 2002; Hughes & Ensor, 2006; Lipscomb et al., 2012). Parental negative affectivity is a predictor of harsh parenting (see Rueger, Katz, Risser, & Lovejoy, 2011), but little is known about mechanisms that may account for this association, how this operates within a mother-father dyadic context, or how these processes may unfold over time. Better understanding of how these two constructs are related over time while children are young offers the opportunity to target maladaptive parenting processes and improve child outcomes. The current study tests whether parental negative affectivity is longitudinally associated with harsh parenting via parenting stress for first-time parents with a child transitioning from late toddlerhood to the early preschool years. We assess this mediational hypothesis within a dyadic context in a sample of couples who participated in a universal couple-based preventive co-parenting education program (Feinberg & Kan, 2008) and examine potential gender differences.

Harsh Parenting during the Transition to Preschool Years

Myriad studies have demonstrated that harsh parenting, defined as a cluster of parental discipline practices involving displays of anger, aggression, and yelling (Arnold, O’Leary, Wolff, Acker, 1993; Hughes & Ensor, 2006; Rueger et al., 2011), is associated with impaired psychological well-being and social functioning in children. For example, children exposed to harsh parenting display increased psychological distress, emotional dysregulation, aggression, delinquency, and antisocial behaviors that can, in turn, increase children’s risk for future maladjustment and conduct problems (e.g., Chang, Schwartz, Dodge, & McBride-Chang, 2003; Gershoff, 2002; Lipscomb et al., 2012). Furthermore, it has been shown that levels of harsh parenting increase from birth to toddlerhood and remain high through the preschool years (e.g., Kim et al., 2010). Given that harsh parenting is often a precursor to child physical abuse (Gershoff, 2002), it is important to identify predictors of harsh parenting during this critical time period to inform prevention and early intervention efforts.

Negative Affectivity and Harsh Parenting

Negative affectivity is a stable and pervasive trait defined as one’s innate and general tendency to experience negative emotions such as distress, discomfort, and dissatisfaction across a variety of contexts (Watson & Clark, 1984). Characterized by the experience of dysphoria, rumination, tension, and agitation (Lonigan, Carey, & Finch, 1994; Watson, Clark & Tellegen, 1988), negative affect is phenomenologically related to symptoms of depression and anxiety, both of which are associated with dysfunctional parenting (e.g., Lovejoy, Graczyk, O’Hare & Neuman, 2000; Van Der Bruggen, Stams & Bogels, 2008). A meta-analytic review has investigated parental negative affectivity as a predictor of harsh parenting practices in non-clinical populations using different operationalizations of parental negative affect, including neuroticism, negative mood, depressive symptoms, and anxiety symptoms. Small-to-medium size associations between negative affectivity and harsh parenting were observed independent of whether negative affectivity was operationalized as an enduring personality trait or a transitory emotional state (Rueger et al., 2011). However, from a prevention perspective, trait negative affect is relatively more stable, and thus potentially a more reliably identified risk factor than state level affect. Thus, better understanding of the underlying mechanism(s) accounting for the association between trait negative affect and harsh parenting may hold important implications for prevention and early intervention efforts. One possible mediator of this relation is the degree of stress new parents experience as they take on the parenting role.

Negative Affectivity and Perceived Stress

Parental negative affectivity is thought to relate to parenting stress because individuals high in negative affect tend to experience high levels of stress across a variety of situations (Caspi, Roberts & Shiner, 2005). Indeed, negative affect has been shown to be positively associated with daily stress (e.g. Gunthert, Cohen, & Amerli, 1999; Schneider, 2004), including parenting stress during the early years of parenthood (e.g., Boyce, Condon, Barton & Corkindale, 2007; Mulsow, Caldera, Pursley, Reifman, & Huston, 2002). People high in negative affect are thought to perceive stressors as threats rather than challenges, which may, in turn, lead to their experiencing higher level of subjective stress (Suls, 2001). For example, there is evidence that negative affectivity is positively associated with perceived stress even after taking into account the objectively considered severity of stressor exposure (Eaton & Bradley, 2008).

The transition to parenthood is a period commonly marked by elevated levels of stress (e.g., Doss, Rhoades, Stanley, & Markman, 2009). Yet, the transition to parenthood is not a discrete event that ends with the birth of the first child or even the first year of parenthood. First-time parents grapple with new challenges in their parenting roles as children’s developmental needs change. The early years of parenthood can be especially difficult, as a child proceeds through multiple developmental stages – all of which require high levels of active caretaking throughout the day and evening – in a relatively short time period. Developmentally, the late toddlerhood and early preschool years are associated with advances in children’s gross motor skills, which permit greater physical independence and assertion of autonomy but still relatively undeveloped emotion regulation skills. With the development of language in early childhood, children are also able to engage in more advanced conversations and to exert their wills verbally. In the context of these challenges, parents higher in negative affectivity may experience more stress in their parenting roles than do parents who are relatively lower in negative affectivity.

Stress and Harsh Parenting

In general, individuals reporting higher levels of stress tend to use emotion-focused coping (e.g., expressing emotions, support seeking) to cope with distress (Eaton & Bradley, 2008). This appears to extend to the parenting domain, as parental stress is positively associated with parents’ use of emotion-focused coping (Dabrowska & Pisula, 2010). One maladaptive form of emotion-focused coping is loss of control and/or hostility towards other people (Connor-Smith & Flachsbart, 2007). Parenting practices such as directing dysregulated negative emotions towards children (e.g., yelling, insulting) may be conceptualized as part of parents’ dysfunctional emotion-focused coping behaviors in response to distress experienced in a parenting context. Indeed, prior research has demonstrated that individuals reporting higher parenting stress tend to incorporate more strict discipline in their parenting practices (Anthony et al., 2005). Thus, parents experiencing higher levels of stress as their child proceeds through the transition to preschool years may engage in greater harsh parenting as one manifestation of maladaptive emotion-focused coping.

Distinctions within Parenting Stress: Personal Distress versus Childrearing Stress

Parenting stress, defined as stress and distress evoked by the parental role and childrearing-related experiences (Abidin, 1995; Haskett, Ahern, Ward, & Allaire, 2006), consists of two related, but distinct, sources of stress related to being a parent: childrearing stress and personal distress (Haskett et al., 2006). That is, stress related to the parenting role involves both distress related to perceiving one’s child as difficult based on direct interaction with the child (e.g., perceiving the child to be moody and easily upset) and distress related to the many constraints imposed by the parenting role on various life domains (e.g., reduced sleep and leisure time due to increased role demands), respectively. Compared to personal distress, childrearing stress may be a stronger predictor of harsh parenting because it is particularly salient to parents when they interact directly with their child. During the interaction with the child, parents may be more likely to act negatively towards the child and/or engage in harsh parenting behaviors to cope with the perceived difficulty of the child at the moment (e.g., to stop a tantruming toddler). Alternatively, personal distress may influence one’s parenting practices by reducing one’s coping capacity across a variety of social contexts. For example, married couples were found to engage in more conflicts with each other on days when either partner reported experiencing high levels of individual stress (Timmons, Arbel, & Margolin, 2016). In a parenting context, it is possible that parents will act more harshly towards their child when they are experiencing high levels of personal distress even if they do not perceive their child to be difficult during the interaction per se. To date, no prior work has directly compared the prospective contributions of these two domains of parenting stress to harsh parenting in a dyadic context during this developmental transition. Greater understanding of potential differences in the extent to which one domain is a more salient predictor of harsh parenting than the other may aid in the development of specific targets for prevention and early intervention efforts to reduce harsh parenting behaviors.

Dyadic Context

Lastly, when examining the longitudinal associations among negative affectivity, parenting stress, and harsh parenting, it is important to consider both members of the mother-father dyad given the linked nature of couples’ affective and behavioral responses and the shared stressor of parenting a child transitioning to the preschool years. From a family systems perspective (Minuchin, 1985), individuals within a family function in an interdependent manner. Thus, with respect to parenting, one parent’s negative affectivity may not only influence his or her own parenting practices (actor effect) but also his or her spouse’s parenting practices (partner effect) as a result of increased negativity, conflict, and stress in the home. Using a dyadic analytic approach also facilitates an explicit test of gender differences in these actor and partner effects. Prior research indicates that, during the early years of child development, women identify more closely with their identity as parents than do men, and they typically spend more time engaging with their child (Cowan & Cowan, 1988; Lang et al., 2014; Stueve & Pleck, 2001). Thus, given the difference in identity salience and parental involvement, the longitudinal association between parental negative affect and harsh parenting through parenting stress may vary as a function of parent gender.

To date, most research examining the relation between parental negative affect and harsh parenting has focused on mothers or fathers, rarely considering the contributions of both parents’ negative affect simultaneously. One exception is a study that employed a dyadic framework to investigate the association between negative affect and parenting practices and found inconclusive evidence for a gender difference in partner effects and no evidence for a gender difference in actor effects (Murdock, Lovejoy, & Oddi, 2014). However, this study did not examine potential mediators (e.g., parenting stress) that could account for this association or examine these associations longitudinally. Thus, the goal of the present investigation was to examine the mediating role of parenting stress – including personal distress and childrearing stress – in the longitudinal pathway from parent negative affect to harsh parenting during early childhood in a dyadic context, along with the exploration of gender differences in those effects.

Hypotheses

Due to the interdependent nature of couples in their parenting role, particularly in the context of a high stress developmental period such as the transition to preschool years, we expected to observe both actor and partner effects of the mediating role of parenting stress in the association between negative affectivity and harsh parenting over time. We hypothesized a longitudinal indirect effect from parental negative affectivity to harsh parenting through parenting stress for both mothers and fathers:

Hypothesis 1: For both mothers and fathers, we hypothesized that negative affectivity would be positively and prospectively associated with one’s own (actor effects; H1a) and one’s partner’s personal distress and childrearing stress (partner effects; H1b). Hypothesis 2: For both mothers and fathers, we hypothesized that personal distress and childrearing stress would each be positively and prospectively associated with one’s own (actor effects; H2a) and one’s partner’s harsh parenting practices (partner effects; H2b). We also explored whether there were differences in the strength of these indirect actor and partner meditational pathways as a function of parent gender and whether there were differences in these pathways as a function of personal distress versus childrearing stress.

Method

Participants and Procedure

Participants for the current study were a subsample of 169 co-resident heterosexual couples recruited for a randomized controlled trial of Family Foundations (FF; Feinberg & Kan, 2008). Five couples from the parent study did not provide any data postpartum, resulting in a sample of 164 couples for the current study. FF is a couple-based psychoeducational transition to parenthood program that helps couples maintain a healthy and strong co-parenting relationship after the birth of their first child and consists of four prenatal and four postnatal sessions (Feinberg & Kan, 2008). Previous research has demonstrated positive intervention effect at post-test on co-parenting, parents’ psychological well-being, and children’s adjustment 6 months postpartum, with effects lasting through 12 months and 3 years postpartum (Feinberg & Kan, 2008; Feinberg, Jones, Kan, & Goslin, 2010; Feinberg, Kan, & Goslin, 2009). Long-term effects on children’s adjustment were also found when children were 5-7 years old (Feinberg, Jones, Roettger, Solmeyer, & Hostetler, 2014). In addition, as relates to the current study, parents in the intervention reported lower levels of parenting stress across the first 3 years postpartum and lower levels of harsh parenting at 3 years postpartum than those in the control group (Feinberg et al., 2010).

Couples were drawn from a mix of rural towns and small cities and were randomly assigned to the FF condition (n = 87) or a control condition in which they were provided with information about how to choose quality childcare (n = 77). For the current study, we used postnatal data collected at three time points: 6 months after birth (Time 1, T1), 12 months after birth (Time 2, T2), and 3 years after birth (Time 3, T3). The attrition rate for mothers and fathers, respectively, was 8% and 7% at T1, 7% and 12% at T2, and 17% and 23% at T3. Recruitment procedures and responses rates for the parent study from which the current sample is drawn can be found in Feinberg & Kan (2008). Inclusion criteria for the parent study required that both participants were 18 years old or older, expecting their first child together, and cohabiting. Upon recruitment, participants’ mean ages were 28.33 (SD = 4.93) years for mothers and 29.76 (SD = 5.58) years for fathers. The majority of participants completed some post-secondary school education (85.6% of mothers and 64.7% of fathers) and identified as European American (91.1% of mothers and 89.9% of fathers). Annual family income ranged from $2,500 to $162,500 with a median of $65,000 (M = $65,076, SD = $34,103). Couples in the control group did not differ from couples in the intervention group with respect to race/ethnicity, education, income, or age (ps > .53). The study was approved by the Institutional Review Board at the Pennsylvania State University, and informed consent was obtained from all participants.

Measures

Negative Affect

For the purposes of the present study, a composite of trait anxiety and depressive symptoms was created to serve as a proxy of parental negative affectivity given negative affectivity’s conceptual overlap with depression and anxiety (Mineka, Watson, & Clark, 1998) and its empirical association with anxiety and depressive symptoms in non-clinical, population-based samples (e.g., Bienvenu et al., 2001; Weinstock & Whisman, 2006). Trait anxiety was assessed at T1 using the 20-item short form of the Taylor Manifest Anxiety Scale (MAS; Bendig, 1956). Sample items included, “I frequently find myself worrying about something” and “I work under a great deal of tension.” Participants responded yes or no to each item. The sum of the 20 items was calculated, with higher scores indicating greater level of trait anxiety. The internal consistency of this measure was good for both mothers and fathers (mothers: α = .85; fathers: α = .79). Depressive symptom severity was assessed at T1 using the Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). Seven items from the original 20 items that demonstrated the highest item-to-total score correlations and accounted for over 90% of the variance in the total score (Howe, Levy & Caplan, 1999) were used. Participants were asked to report about their symptoms in the past week (e.g., “How often during the past week did you feel bothered by things that usually don’t bother you?” “How often during the past week did you feel your life had been a failure?”) on a four-point scale (0 = Rarely or None of the Time, 4 = Most or All of the Time), with higher mean scores indicating higher levels of depressive symptoms. The internal consistency of this measure was adequate for both mothers and fathers (mothers: α = .79; fathers: α = .76). Bivariate correlations within individuals indicated that parental anxiety and depression measured at T1 were strongly correlated (r_fathers = .57; r_mothers = .67), and the composite variable was created by taking the mean of the two standardized variables.¹

Parenting Stress

Parenting stress was assessed at T2 using the Parenting Stress Index Short Form (PSI-SF; Abidin, 1995). Twenty-seven of the original 36 items were used instead of the full 36-item scale, as nine items were found to have low factor loadings relative to the other items (Abidin, 1995). Parenting stress was measured with 12 items on personal distress after taking on the role as a parent (e.g., “I find myself giving up more of my life to meet my children’s needs than I ever expected;” “I don’t enjoy things as I used to;” “Since having this child, I have been unable to do new and different things”; mothers: α = .82; fathers: α = .80) and 15 items on childrearing stress (e.g., “My child makes more demands on me than most children;” “I feel that my child is very moody and easily upset”; mothers: α = .86; fathers: α = .80). Each item is scored on a 5-point Likert scale (1 = Strongly Agree, 5 = Strongly Disagree). All items were reverse coded, with higher mean scores indicating greater parenting stress.

Harsh Parenting

Harsh parenting was assessed at T3 using nine items from The Parenting Scale (Arnold et al., 1993). Each item (e.g. “When my child misbehaves…”) is scored on a scale from 1 (e.g., “I speak to my child calmly”) to 7 (e.g., “I raise my voice or yell”). Higher mean scores reflect higher levels of harsh parenting. The internal consistency of this measure was adequate for both mothers and fathers (mothers: α = .76; fathers: α = .78).

Analytic Plan

To test our hypothesized mediational model that parental negative affect would prospectively predict harsh parenting through personal distress and childrearing stress in a dyadic context, we utilized Actor-Partner Interdependence Mediation Model (APIMeM; Ledermann, Macho, & Kenny, 2011) within a Structural Equation Modeling framework using Mplus7 (Muthén & Muthén, 1998–2012). The Actor Partner Interdependence Model (APIM; Kenny, Kashy, & Cook, 2006) allows for the testing of how members of a dyad (i.e., mother and father) influence each other (partner effect) after taking into account one’s own influence on an outcome (actor effect) and the interdependence between the two members in the dyad. As an extension of APIM, APIMeM allows for the testing of mediation in a dyadic context. Consistent with the non-clinical nature of the sample, the negative affect variable was positively skewed. Thus, a maximum likelihood estimator with robust standard errors (MLR) was used as the method of estimation. Good model fit was interpreted according to Hu and Bentler’s (1999) recommendations: (a) a small and non-significant chi-square test; (b) the root mean square error of approximation (RMSEA) less than or equal to .06; (c) the root mean squared residual (SRMR) less than .08; and (d) the comparative fit index (CFI) and the Tucker-Lewis index (TLI) greater than or equal to .95.

Consistent with Rucker, Preacher, Tormala & Petty’s (2011) assertion that a significant X → Y effect is not a requirement prior to examining indirect effects (X → M → Y) and that one should instead focus on evaluating the size and significance of indirect effects, we directly tested and compared mediation by the two proposed mediators (i.e., personal distress and childrearing stress) within the APIMeM model. As displayed in Figure 1, T2 personal distress and childrearing stress were examined as two simultaneous mediators of the association between T1 negative affect and T3 harsh parenting. To obtain the most parsimonious model, a step-wise modeling procedure was utilized whereby chi-square difference tests (adjusted for the use of MLR) were used to make model comparisons in a sequential fashion, with the more parsimonious models being retained when the fit of the model with fewer paths estimated did not degrade significantly. First, a fully saturated model in which all paths were freely estimated was tested. Second, moderation by gender was tested by comparing models in which the paths were constrained to be the same versus freely estimated across fathers and mothers. Third, the strength of the association between negative affect and personal distress was compared with the strength of the association between negative affect and childrearing stress. Fourth, the strength of the association between personal distress and harsh parenting was compared with the strength of the association between childrearing stress and harsh parenting. Lastly, bootstrapping was used to directly evaluate mediating mechanisms within the APIMeM framework with 5000 samples (Preacher & Hayes, 2008) to test the significance of the indirect effects of T1 negative affect on T3 harsh parenting through T2 personal distress and childrearing stress for mothers and fathers. The strength of the indirect effects through both domains of stress (i.e., personal distress vs childrearing stress) was then compared across gender and finally, within gender, was compared as a function of personal distress versus childrearing stress.

Figure 1.

The mediation model with standardized parameters depicting the association between negative affect 6 months postpartum and harsh parenting 3 years postpartum via personal distress and childrearing stress 12 months postpartum. N = 164 couples.

Note. X = exogenous variable; M = mediator; Y = endogenous variable; m = mothers; f = fathers. ** p < .01. *** p < .001. For ease of presentation, parameter estimates for non-significant paths (dashed lines), correlations between predictor variables and residual covariances as well as the paths for control variables are not displayed but are available from the first author upon request.

Intervention status (0 = control, 1 = intervention) was entered as a control variable in all models given the intervention effect on postnatal parenting stress and harsh parenting (Feinberg et al., 2010).² In light of documented associations between child difficulty and parenting practices in other studies (e.g., Koenig, Barry, & Kochanska, 2010), composite measures of observational ratings of child difficulty assessed at T2 and T3 were also controlled at T2 and T3, respectively.³ Race, age, educational background, and household income were initially included as covariates.⁴ For ease of presentation, all covariates other than intervention status were removed from the final models as the pattern of findings did not differ.

Results

Table 1 presents the descriptive statistics and bivariate correlations among mothers and fathers’ T1 negative affect, T2 personal distress and childrearing stress, and T3 harsh parenting.

Table 1.

Correlations, Means and Standard Deviations for Mother-Father Dyads (N = 121 to 156)

	1	2	3	4	5	6	7	8
1. Mother’s Negative Affect	–
2. Father’s Negative Affect	.11	–
3. Mother’s Personal Distress	.49**	.11	–
4. Father’s Personal Distress	.27**	.37**	.29**	–
5. Mother’s Childrearing Stress	.27**	−.07	.57**	.18*	–
6. Father’s Childrearing Stress	.30**	.10	.25**	.52**	.28**	–
7. Mother’s Harsh Parenting	.24**	−.03	.21**	.17	.24**	.21*	–
8. Father’s Harsh Parenting	.13	.14	−.01	.31**	.01	.25**	.30**	–
M	.00	.00	2.17	2.16	1.70	1.79	2.38	2.43
SD	.91	.88	.58	.53	.49	.41	.68	.76

Note.

^*p < .05;

^**p < .01.

The Longitudinal Actor Partner Mediation Model

Figure 1 presents the final APIMeM model with significant standardized estimates for T1 parental negative affect predicting T2 parenting stress (personal distress and childrearing stress), which, in turn, predicts T3 harsh parenting. The non-significant adjusted chi-square, along with other model fit indices, suggested that the model fit the data well: χ²(15) = 17.07, p = .31; RMSEA = .03, 90% confidence interval (CI) [.00, .08]; SRMR = .056; CFI = .99; and TLI = .98.

Mediation by Personal Distress and Childrearing Stress

Personal Distress

As displayed in Figure 1, for both mothers and fathers, there were significant actor (one’s characteristic on one’s outcome) and partner effects (one’s characteristic on one’s partner’s outcome). Specifically, there were positive, medium-sized actor effects from T1 negative affect to T2 personal distress, positive, small-sized actor effects from T2 personal distress to T3 harsh parenting, and positive, small-sized partner effects from T1 negative affect to T2 personal distress for both mothers and fathers. There was no evidence of partner effects from T2 personal distress to T3 harsh parenting.

Childrearing Stress

There were positive, small-sized actor effects from T1 negative affect to T2 childrearing stress and from T2 childrearing stress to T3 harsh parenting for both mothers and fathers. There were also positive, small-sized partner effects for mothers only (i.e., from mother’s T1 negative affect to father’s T2 childrearing stress). There was no evidence of partner effects from father’s T1 negative affect to mother’s T2 childrearing stress or partner effects from T2 childrearing stress to T3 harsh parenting for mothers or fathers.

Moderation by Gender

Moderation by gender was tested using chi-square difference tests comparing models in which significant paths (i.e., from T1 negative affect to T2 parenting stress, from T2 parenting stress to T3 harsh parenting) were freely estimated across gender to models in which paths were constrained to be the same across gender. There was a gender difference in the partner effect from one’s T1 negative affect to one’s partner’s T2 childrearing stress (Δχ²(1) = 8.92, p = .003) such that mothers’ T1 negative affect positively and significantly predicted fathers’ T2 childrearing stress but not vice versa. There was no gender difference in the partner effects from T1 negative affect to one’s partner’s T2 personal distress (Δχ²(1) = 1.93, p = .16), nor were there gender differences in the actor effects from T1 negative affect to T2 personal distress (Δχ²(1) = 1.90, p = .17), from T1 negative affect to T2 childrearing stress (Δχ²(1) = 2.83, p = .09), from T2 personal distress to T3 harsh parenting (Δχ²(1) = 3.03, p = .08), or from T2 childrearing stress to T3 harsh parenting (Δχ²(1) = 1.15, p = .28).

Personal Distress versus Childrearing Stress

The chi-square difference test was then used to test whether the size of the path from T1 negative affect to T2 parenting stress and from T2 parenting stress to T3 harsh parenting differed as a function of personal distress versus childrearing stress (actor effects). For both fathers and mothers, the strength of the association between one’s T1 negative affect and one’s T2 personal distress was significantly stronger than the association between one’s T1 negative affect and one’s T2 childrearing stress (Δχ²(1) = 20.26, p < .001). There was no difference in actor effects from T2 parenting stress to T3 harsh parenting as a function of parenting stress domain (Δχ²(1) = .02, p = .89). Lastly, the magnitude of the three significant partner effects from T1 negative affect to T2 parenting stress (i.e., mothers’ T1 negative affect to father’s T2 personal distress, fathers’ T1 negative affect to mothers’ T2 personal distress, and mothers’ T1 negative affect to father’s T2 childrearing stress) did not differ (Δχ²(1) = .23, p = .63).

Indirect Effects of Negative Affect on Harsh Parenting

Given the actor and partner effects reported above, indirect actor effects (i.e., the effects from one’s T1 negative affect to one’s T3 harsh parenting through one’s own T2 personal distress and childrearing stress) and indirect partner effects (i.e., the effects from one’s T1 negative affect to one’s partner’s T3 harsh parenting through one’s partner’s T2 personal distress and childrearing stress and the effects from one’s T1 negative affect to one’s own T3 harsh parenting through one’s partner’s T2 personal distress and childrearing stress) were directly tested using bootstrapping procedures. Unstandardized estimates of those indirect effects, p-values, and 95% bias-corrected bootstrap confidence intervals calculated with 5000 bootstrap samples are presented in Table 2.

Table 2.

Indirect Effects from the Actor–Partner Interdependence Mediation Model with Parental Negative Affect as Exogenous Variable, Parenting Stress (Personal Distress & Childrearing Stress) as Mediator, and Harsh Parenting as Endogenous Variable

Effect		IE	95% CI	p
Personal Distress as the Mediator
Xm → M1m →Ym	= A → A	.055	.028,.087	<.001
Xf → M1f →Yf	= A → A	.055	.028,.087	<.001
Xm → M1f →Ym	= P → P	−.001	−.013,.012	.814
Xf → M1m →Yf	= P → P	−.001	−.013,.012	.814
Xm → M1m →Yf	= A → P	−.003	−.033,.025	.817
Xf → M1f →Ym	= A → P	−.003	−.033,.025	.817
Xm → M1f →Yf	= P → A	.023	.009,.045	.007
Xf → M1m →Ym	= P → A	.023	.009,.045	.007
Childrearing Stress as the Mediator
Xm → M2m →Ym	= A → A	.020	.005,.041	.018
Xf → M2f →Yf	= A → A	.020	.005,.041	.018
Xm → M2f →Ym*	= P → P	−.001	−.013,.012	.814
Xf → M2m →Yf*	= P → P	.000	−.005,.005	.969
Xm → M2m →Yf	= A → P	−.001	−.015,.008	.820
Xf → M2f →Ym	= A → P	−.001	−.015,.008	.820
Xm → M2f →Yf*	= P → A	.023	.009,.045	.007
Xf → M2m →Ym*	= P → A	.000	−.017,.016	.968

Note. Table values are unstandardized coefficients. IE = indirect effect; CI = confidence interval; X = exogenous variable; M = mediator; Y = endogenous variable; f = father; m = mother; M1 = personal distress; M2 = childrearing stress; A = actor effect; P = partner effect. For example, X_m → M1_m →Y_m represents the mediating effects from mothers’ negative affect to mothers’ harsh parenting through mothers’ personal distress. A → A represents the actor effect from negative affect to parenting stress and the actor effect from parenting stress to harsh parenting. All the indirect effects were constrained to be the same across gender except for the indirect effects marked by asterisks (^*).

For fathers and mothers, the indirect actor effects from T1 negative affect to T3 harsh parenting through T2 personal distress (Xm/f→M1m/f→Ym/f in Figure 1) and T2 childrearing stress (Xm/f→M2m/f→Ym/f in Figure 1) were both significant, with a stronger mediating effect for personal distress than for childrearing stress (p = .003). The indirect partner effects from T1 negative affect to partner’s T3 harsh parenting were also significant for mothers and fathers, with a stronger indirect effect from mothers to fathers than from fathers to mothers (p = .01). Specifically, the indirect partner effect from mothers’ negative affect to fathers’ harsh parenting was mediated through both fathers’ personal distress (Xm→M1f→Yf in Figure 1) and childrearing stress (Xm→M2f→Yf in Figure 1) whereas the indirect partner effect from fathers’ T1 negative affect to mothers’ T3 harsh parenting was only mediated through mothers’ personal distress (Xf→M2m→Ym in Figure 1).

Discussion

The goal of the current study was to examine whether parental negative affect at 6 months postpartum (T1) prospectively predicted harsh parenting at 3 years postpartum (T3) via parenting stress one year postpartum (T2) in a dyadic context (i.e., using reports from both parents) for first-time parents with a child transitioning from late toddlerhood to the early preschool years. We also examined two domains of parenting stress – personal distress and childrearing stress – to determine if there were differences in the mediating effects of these two sources of parenting stress and examined potential gender differences in these associations.

In summary, we found evidence for indirect actor effects as hypothesized for both mothers and fathers: both domains of parenting stress mediated the longitudinal association between parental negative affectivity and the same parent’s harsh parenting. Although not hypothesized, we observed a stronger mediation effect for personal distress than childrearing stress. In addition, there was evidence for indirect partner effects from parental negative affectivity to the other partner’s harsh parenting for both mothers and fathers. This indirect partner effect was stronger for mothers’ negative affect predicting fathers’ harsh parenting, which was mediated through both domains of parenting stress, than for father’s negative affect predicting mother’s parenting, which is mediated through mother’s personal distress only.

Of note, the pattern of findings in the current study held whether or not covariates reflecting general external life stress (e.g., socioeconomic status) and observer-rated difficulty of the child were included in the model. That is, parental negative affect prospectively predicted harsh parenting behaviors through both domains of parenting stress even after controlling for objectively considered severity of stressors outside of the individual. These findings are consistent with earlier work that has identified parental characteristics as salient predictors of parenting stress above and beyond child characteristics (Belsky, 1984; Mulsow et al., 2002).

The actor effects that we observed for the associations between negative affect and parenting stress, as well as between parenting stress and harsh parenting, for both mothers and fathers replicate those observed in previous studies (e.g., Anthony et al., 2005; Boyce et al., 2007; Mulsow et al., 2002; Ponnet et al., 2013); but, our findings also extend our understanding of the association between negative affect and harsh parenting by demonstrating longitudinal mediating effects. This finding may help to explain the intervention effects observed in the parent study on postnatal parenting stress and harsh parenting; specifically, parents in the intervention engaged in less harsh parenting behaviors compared to the control group, and this may be due to the lower levels of parenting stress they experience across the first 3 years postpartum. With regard to the stronger mediating effect of personal distress than childrearing stress, we found that parental negative affect was more strongly predictive of personal distress than childrearing stress; in contrast, personal distress and childrearing stress prospectively predicted harsh parenting in a similar fashion. One reason that negative affect may be more strongly predictive of personal distress than childrearing stress is that personal distress represents the accumulated impact of the parenting role on multiple life domains (e.g., work, finances, leisure) whereas childrearing stress centers primary on one’s direct interactions with one’s child. However, it may also be that some of the items assessed in the personal distress domain (e.g., “I don’t enjoy things as I used to”) share greater conceptual overlap with parental negative affect than do the items in the childrearing stress domain.

At the dyadic level, partner effects were modeled in light of the theorized and demonstrated interdependence between mothers and fathers in their parenting role (Barnett, Deng, Mills-Koonce, Willoughby, & Cox, 2008; Minuchin, 1985). To our knowledge, only one study has examined the association between negative affect and harsh parenting in a dyadic context (Murdock et al., 2014). The investigators found a partner effect from mothers’ negative affect to fathers’ harsh parenting but did not explicitly test mediators of this association. Our investigation extended these findings by documenting an indirect partner effect from parent’s negative affect to the other partner’s harsh parenting through partner’s parenting stress, specifically through partner’s personal distress for both men and women and through father’s childrearing stress from mothers’ negative affect to fathers’ harsh parenting only. Moreover, our results indicate that the effects of negative affect on a partner’s parenting behaviors are the result of the influence of negative affect on the partner’s parenting stress, rather than the influence of one’s parenting stress on the other partner’s harsh parenting. That is, an individual’s parenting stress is influenced by both one’s own and one’s partner’s negative affect, whereas one’s parenting behaviors appear to be accounted for primarily by one’s own stress.

One reason for this set of findings could be that partners of individuals high in negative affect experience greater parenting stress as a result of constraints and burden imposed on them by virtue of having to accommodate their partners’ diminished functioning in both the parenting and co-parenting roles. This increased parental stress subsequently increases one’s risk of engaging in harsh parenting behaviors, which in the current study was conceptualized as a dysfunctional emotion-focused coping strategy (Eaton & Bradley, 2008). In contrast, living with a partner experiencing high levels of parenting stress does not appear to influence one’s own harsh parenting. Consistent with this assertion, Ponnet et al. (2013) examined the influence of childrearing stress on demandingness in parenting, a construct related to harsh parenting, within a dyadic context and found evidence for actor effects for childrearing stress on demandingness but no evidence for partner effects. Thus, although one’s stress level may be amplified by one’s partner’s negative affect, the behavior that one engages in to cope with emotional distress seems to be an intra-individual process that is governed by the internal stress level one experiences.

Our findings also revealed a gender difference in the indirect partner effect from negative affect to harsh parenting. Specifically, the influence of mothers’ negative affect on fathers’ harsh parenting through childrearing stress was stronger than vice versa. That is, although both mothers and fathers’ personal distress are influenced by each other’s negative affect in a similar fashion, fathers’ childrearing stress is influenced by mothers’ negative affect but not the reverse. The gender difference in this partner effect could be due to the difference in men and women’s perceived parenting competence. Childrearing stress is defined as stress related to perceived child difficulty during direct interaction with the child. In Western societies, women tend to take greater responsibility for parenting tasks (Dempsey, 2002) and are commonly thought to have greater parenting competence than men (Cowan & Cowan, 1988). During the early years of child development, fathers in heterosexual relationships tend to perceive themselves as assisting their partners in the caretaking role rather than taking primary or co-equal responsibility for child care (Stueve & Pleck, 2001). Previous research has found evidence indicating that fathers’ parenting behaviors are more susceptible to the influence of their partner’s co-parenting support than mothers’ parenting behaviors (Pedro, Ribeiro, & Shelton, 2012). Thus, fathers partnered with women high in negative affect may experience more childrearing stress in their direct interactions with the child as a result of their partners’ (i.e., mothers) diminished functioning as a supportive co-parent. By contrast, mothers’ childrearing stress may not be influenced by their partners’ negative affect because they take primary responsibility for caregiving and, thus, are less reliant on spousal support in their parenting role.

Implications for Prevention and Intervention Efforts

Findings from the present study suggest potential targets for future prevention and early intervention efforts designed to reduce or prevent maladaptive discipline parenting practices. Both parental negative affect and its influence on parenting stress may be conceptualized as risk factors for both mothers’ and fathers’ harsh parenting. As such, identifying couples in which one or both members are high in negative affect early in their transition to parenthood and providing them with resources to reduce parenting stress may be a promising strategy for preventing harsh parenting. Efforts to decrease parenting stress could consist of helping to reduce the external stressors in the parents’ life, including the enhancement of food and housing security for poor families or the provision of high-quality and accessible childcare. Such efforts could also include attenuating the influence of negative affect on parenting stress by helping parents high in negative affect learn to develop more adaptive attributions for perceived difficulties with children, more adaptive appraisals of their own parenting capacities, and greater engagement in personal self-care. In addition to teaching strategies that directly target parenting stress, offering couple-based interventions that promote emotion regulation skills for individuals prone to negative affect (Kirby & Baucom, 2007) may also help to reduce one’s own and one’s partner’s harsh parenting by facilitating partners’ co-regulation of emotions in a constructive manner. Thus, for couples deemed at high risk by virtue of experiencing negative affectivity in one or both partners, the inclusion of strategies to reduce external stressors as well as dyadic coping strategies in response to individual and dyadic level stress may be useful in augmenting the effects of universal, couple-based parenting programs.

Limitations and Future Directions

There are several limitations associated with the current study. First, this study relied exclusively on men and women’s self-report of negative affectivity, parenting stress, and harsh parenting. Future studies that incorporate other methods of measurement of individuals’ stress levels would be helpful in reducing shared method variance. For example, fundamental frequency, an index of vocal tension, has been used as an indicator of emotional arousal (Weusthoff, Baucom, & Hahlweg, 2013). Second, our sample was relatively homogeneous with respect to ethnicity and education in that most of the sample was Caucasian and relatively well educated. In addition, our sample was relatively high functioning, with a somewhat restricted range of harsh parenting and parenting stress. Future studies on this topic would benefit from a more diverse sample with respect to demographic characteristics and family functioning to determine whether the pattern of findings observed in this sample generalize to other populations. Third, although the association between parental negative affect and harsh parenting does not differ whether state or trait negative affect is assessed (Rueger et al., 2011), we recognize that it would have been more straightforward and, therefore, preferable to measure negative affect in this study using a more direct and specific measure of trait negative affect, such as neuroticism. Future studies should attempt to replicate these longitudinal associations by using neuroticism or other measures of trait negative affect. Fourth, although utilizing longitudinal, temporally ordered data increases confidence in the directionality of effects, our analyses cannot definitively determine whether these are causal effects. We were not able to test and compare alternative models in which the longitudinal associations among the study variables are in the opposite direction than what was proposed (e.g., parenting stress leads to parental negative affect) and the possibility of the bidirectional associations among the study variables. Future studies should assess all study variables with repeated measures. In addition, there were positive intervention effects demonstrated in parenting stress across first 3 years postpartum and harsh parenting assessed at 3 years postpartum (Feinberg et al., 2010). Although our results suggest that the intervention may exert its effect on harsh parenting by reducing parental stress, the intervention could also have exerted its effect on parenting stress and harsh parenting by attenuating associations among the study variables. Moderation by intervention status was not supported in these paths; however, this could have been due to low statistical power given the complexity of the model. With a larger sample size, future studies may be better positioned to characterize intervention effects on the strength of the association among the study variables. Lastly, there may be selection bias with the current sample, as couples in this study self-selected into a randomized controlled trial of a couple-based psychoeducational transition to parenthood program and those assigned to the control group stayed in the study long enough for the follow up data collection. Thus, replication of these findings in naturalistic samples would be useful in determining the generalizability of these findings outside the context of intervention studies.

The current study adds to the early parenting literature by demonstrating the longitudinal indirect effects of one’s own and one’s partner’s parental negative affect on harsh parenting 2.5 years later through one’s own parenting stress. These findings highlight the importance of adopting a dyadic analytic framework and suggest the value of taking a couple-based approach for psychoeducational transition to parenthood programs. Among couples in which at least one member of the dyad experiences negative affectivity, inclusion of dyadic coping strategies to mitigate the effects of one’s own and one’s partner’s parenting stress may be beneficial in reducing both partners’ harsh parenting among first-time parents with children transitioning from toddlerhood to the preschool years. In turn, this may prevent escalation from harsh parenting to child abuse and help reduce children’s risk for impaired psychological well-being and social functioning in the future.